perf: optimised source range resolving

Author: NickSdot

src/Source/File.php

@@ -36,30 +36,9 @@ public function lines(): \Generator
     }
 
     /** @throws \OutOfBoundsException if the offset lies outside the source */
-    public function lineAtOffset(int $offset): Line
+    public function lineNumberAtOffset(int $offset): int
     {
-        $sourceLength = strlen($this->content);
-        if ($offset < 0 || $offset > $sourceLength) {
-            throw new \OutOfBoundsException(
-                sprintf('Source offset %d is outside the valid range 0..%d.', $offset, $sourceLength),
-            );
-        }
-
-        $lineBeginOffsets = $this->lineBeginOffsets();
-        $low = 0;
-        $high = count($lineBeginOffsets) - 1;
-
-        while ($low < $high) {
-            $middle = intdiv($low + $high + 1, 2);
-
-            if ($lineBeginOffsets[$middle] <= $offset) {
-                $low = $middle;
-            } else {
-                $high = $middle - 1;
-            }
-        }
-
-        return $this->createLineAtOffset($low + 1, $lineBeginOffsets[$low]);
+        return $this->lineIndexAtOffset($offset, $this->lineBeginOffsets()) + 1;
     }
 
     public function withContent(string $content): self
@@ -77,18 +56,58 @@ private function lineBeginOffsets(): array
         }
 
         $lineBeginOffsets = [0];
+        $sourceLength = strlen($this->content);
+        $lineBeginOffset = 0;
+
+        while ($lineBeginOffset < $sourceLength) {
+            $lineLength = strcspn($this->content, "\r\n", $lineBeginOffset);
+            $lineEndingOffset = $lineBeginOffset + $lineLength;
 
-        foreach ($this->lines() as $line) {
-            if ($line->number === 1) {
-                continue;
+            if ($lineEndingOffset === $sourceLength) {
+                break;
             }
 
-            $lineBeginOffsets[] = $line->beginOffset;
+            $lineBeginOffset = $lineEndingOffset + (
+                $this->content[$lineEndingOffset] === "\r"
+                && ($this->content[$lineEndingOffset + 1] ?? null) === "\n"
+                    ? 2
+                    : 1
+            );
+            $lineBeginOffsets[] = $lineBeginOffset;
         }
 
         return $this->lineBeginOffsets = $lineBeginOffsets;
     }
 
+    /**
+     * @param non-empty-list<int> $lineBeginOffsets
+     * @throws \OutOfBoundsException if the offset lies outside the source
+     */
+    private function lineIndexAtOffset(int $offset, array $lineBeginOffsets): int
+    {
+        $sourceLength = strlen($this->content);
+        if ($offset < 0 || $offset > $sourceLength) {
+            throw new \OutOfBoundsException(
+                sprintf('Source offset %d is outside the valid range 0..%d.', $offset, $sourceLength),
+            );
+        }
+
+        $low = 0;
+        $high = count($lineBeginOffsets) - 1;
+
+        while ($low < $high) {
+            $middle = intdiv($low + $high + 1, 2);
+
+            if ($lineBeginOffsets[$middle] <= $offset) {
+                $low = $middle;
+            } else {
+                $high = $middle - 1;
+            }
+        }
+
+        return $low;
+    }
+
     private function createLineAtOffset(int $lineNumber, int $lineBeginOffset): Line
     {
         $sourceLength = strlen($this->content);

src/Violation/SourceRange.php

@@ -15,7 +15,7 @@
     public static function fromFile(File $file, int $beginOffset, int $untilOffset): self
     {
         return new self(
-            line: $file->lineAtOffset($beginOffset)->number,
+            line: $file->lineNumberAtOffset($beginOffset),
             beginOffset: $beginOffset,
             untilOffset: $untilOffset,
             content: substr($file->content, $beginOffset, $untilOffset - $beginOffset),

tests/Unit/Fix/WhitespaceFixerTest.php

@@ -12,7 +12,6 @@
 use DocbookCS\Runner\RunMode;
 use DocbookCS\Sniff\WhitespaceSniff;
 use DocbookCS\Source\File;
-use DocbookCS\Source\Line;
 use DocbookCS\Violation\SourceRange;
 use DocbookCS\Violation\Violation;
 use PHPUnit\Framework\Attributes\CoversClass;
@@ -31,7 +30,6 @@
     //
     UsesClass(File::class),
     UsesClass(FixPlan::class),
-    UsesClass(Line::class),
     UsesClass(SourceRange::class),
 ]
 final class WhitespaceFixerTest extends TestCase

tests/Unit/Sniff/WhitespaceSniffTest.php

@@ -6,7 +6,6 @@
 
 use DocbookCS\Sniff\WhitespaceSniff;
 use DocbookCS\Source\File;
-use DocbookCS\Source\Line;
 use DocbookCS\Violation\SourceRange;
 use DocbookCS\Violation\Violation;
 use PHPUnit\Framework\Attributes\CoversClass;
@@ -19,7 +18,6 @@
     CoversClass(WhitespaceSniff::class),
     //
     UsesClass(File::class),
-    UsesClass(Line::class),
     UsesClass(SourceRange::class),
 ]
 final class WhitespaceSniffTest extends TestCase

tests/Unit/Source/FileTest.php

@@ -39,17 +39,17 @@ public function itRepresentsLinesAndTheirEndings(): void
     }
 
     #[Test]
-    public function itResolvesOffsetsToLines(): void
+    public function itResolvesOffsetsToLineNumbers(): void
     {
         $file = new File('file.xml', "one\r\ntwo\nthree");
 
-        self::assertSame(1, $file->lineAtOffset(0)->number);
-        self::assertSame(1, $file->lineAtOffset(3)->number);
-        self::assertSame(1, $file->lineAtOffset(4)->number);
-        self::assertSame(2, $file->lineAtOffset(5)->number);
-        self::assertSame(2, $file->lineAtOffset(8)->number);
-        self::assertSame(3, $file->lineAtOffset(9)->number);
-        self::assertSame(3, $file->lineAtOffset(14)->number);
+        self::assertSame(1, $file->lineNumberAtOffset(0));
+        self::assertSame(1, $file->lineNumberAtOffset(3));
+        self::assertSame(1, $file->lineNumberAtOffset(4));
+        self::assertSame(2, $file->lineNumberAtOffset(5));
+        self::assertSame(2, $file->lineNumberAtOffset(8));
+        self::assertSame(3, $file->lineNumberAtOffset(9));
+        self::assertSame(3, $file->lineNumberAtOffset(14));
     }
 
     #[Test]
@@ -62,7 +62,7 @@ public function itIncludesAnEmptyLineAfterTheFinalLineEnding(): void
             new Line(2, '', '', 4),
         ], iterator_to_array($file->lines()));
 
-        self::assertSame(2, $file->lineAtOffset(4)->number);
+        self::assertSame(2, $file->lineNumberAtOffset(4));
     }
 
     #[Test]
@@ -91,6 +91,6 @@ public function itRejectsOffsetsOutsideTheSource(): void
         $file = new File('file.xml', 'one');
 
         $this->expectException(\OutOfBoundsException::class);
-        $file->lineAtOffset(4);
+        $file->lineNumberAtOffset(4);
     }
 }

tests/Unit/Violation/AffectedRangesTest.php

@@ -5,7 +5,6 @@
 namespace DocbookCS\Tests\Unit\Violation;
 
 use DocbookCS\Source\File;
-use DocbookCS\Source\Line;
 use DocbookCS\Violation\Severity;
 use DocbookCS\Violation\SourceRange;
 use DocbookCS\Violation\Violation;
@@ -20,7 +19,6 @@
     CoversClass(Violation::class),
     //
     UsesClass(File::class),
-    UsesClass(Line::class),
 ]
 final class AffectedRangesTest extends TestCase
 {